Hi #mldotnet #dotnet and #onnx people.
Would anyone have any ideas why an ML.NET transformer (pipeline consists of just TokenizeIntoCharactersAsKeys, MapValueToKey, and ProduceNgrams) would throw a "The targeted pipeline can not be fully converted into a well-defined ONNX model." exception?
According to https://learn.microsoft.com/en-us/dotnet/machine-learning/resources/transforms all those transforms are ONNX exportable.
(Working on cleaning up code for a branch to demonstrate the issue. I'll link the branch in a reply when it's done.)